[Retros] comments on Francois' Labelle webpage

Andrew Buchanan andrew at anselan.com
Sat Jan 24 13:01:02 EST 2004

Francois,

Your webpage
http://www.cs.berkeley.edu/~flab/chess/statistics-positions.html looks
great. A couple of points on terminology.

(1) "position"

> As stated on the chess game page, I am ignoring the "draw by repetition of

position" rule. If not, the

> chess position would have to contain information about every move since

the last "irreversible" move (a

> capture or a pawn move).

Actually, your definition of position is exactly that specified in the "draw
by repetition rule". Think about it: if "position" included details of every
move, it could never be repeated! The *consequences* of draw by repetition
and by 50 moves can reasonably be ignored however since they are optional
anyway.

"Position" is used in varying ways by problemists, but for your purposes you
need to pin it down. I suggest you just state the Laws definition, and use
that. In a way, problemists have no right to redefine that term, since it's
in the Laws. Yet we do.

(2) "position" vs "diagram"

> A chess diagram is a chess position with no information on availability of

castling and en passant

> captures, just the contents of the 64 squares. The difference is subtle,

but sometimes important.

> Diagrams are preferred by the chess problems community, and for these

people "position" will always

> mean "diagram".

Generally speaking, chess problemists prefer the pieces to do the talking,
rather than add extraneous words to a problem. This is particularly true for
retrograde analysis, where the idea is to be able to *deduce* interesting
features such as who has the move, castling or ep. And if RA does not give
all the answers, there are conventions about these matters for orthodox
problems. This is an aesthetic thing: "Look Ma no hands!"

Next note that a "diagram" also omits info about who has the move. But this
is automatically supplied again with the stipulation of an SPG.

Neither "position" nor "diagram" has a standard definition for problemists,
so it's too strong to say as you did: 'Diagrams are preferred by the chess
problems community, and for these people "position" will always mean
"diagram".'

We've covered "position" above. The status of the term "diagram" is
interesting. It does have a precise meaning, but the term is just not used
by problemists to describe "a position minus any non-visual information".
They don't talk about "How many diagrams are there at 3.5 moves?" But why
not? As you pointed out, this concept is exactly what we need! So let's take
this term which is just waiting to be used.

All your statistics can perfectly well be presented at the level of
positions. As you point out, that's the canonical representation, at which
recursion operates. It's just when you come to present composed proof games
"Francois+computer" that it's perhaps more aesthetic to collapse the
positions into diagrams.

Anyway cheers for now.
Andrew.

More information about the Retros mailing list